SGM3D: Stereo Guided Monocular 3D Object Detection

Zhou, Zheyuan; Liang, Dingkong; Ye, Xiaoqing; Zou, Zhikang; Tan, Xiao; Zhang, Li; Xue, Xiangyang; Feng, Jianfeng

doi:10.1109/lra.2022.3191849

Cited by 21 publications

(4 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[213] for environment-specific object detection , ResNet [232] trained by Places365 [233] for scene classification. Authors in [234] use KITTI [160] and Lyft [235] dataset to train a doubled stereo-guided monocular 3D (SGM3D) object detection framework based on monocular images for autonomous vehicles navigation.…”

Section: Cameramentioning

confidence: 99%

Recent Advancements in Deep Learning Applications and Methods for Autonomous Navigation -- A Comprehensive Review

Golroudbari¹,

Sabour²

2023

Preprint

View full text Add to dashboard Cite

This review paper presents a comprehensive overview of end-to-end deep learning frameworks used in the context of autonomous navigation, including obstacle detection, scene perception, path planning, and control. The paper aims to bridge the gap between autonomous navigation and deep learning by analyzing recent research studies and evaluating the implementation and testing of deep learning methods. It emphasizes the importance of navigation for mobile robots, autonomous vehicles, and unmanned aerial vehicles, while also acknowledging the challenges due to environmental complexity, uncertainty, obstacles, dynamic environments, and the need to plan paths for multiple agents. The review highlights the rapid growth of deep learning in engineering data science and its development of innovative navigation methods. It discusses recent interdisciplinary work related to this field and provides a brief perspective on the limitations, challenges, and potential areas of growth for deep learning methods in autonomous navigation. Finally, the paper summarizes the findings and practices at different stages, correlating existing and future methods, their applicability, scalability, and limitations. The review provides a valuable resource for researchers and practitioners working in the field of autonomous navigation and deep learning.

show abstract

Section: Cameramentioning

confidence: 99%

Recent Advancements in Deep Learning Applications and Methods for Autonomous Navigation -- A Comprehensive Review

Golroudbari¹,

Sabour²

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Due to various data modalities that can be leveraged during training such as images, videos, and LiDAR point clouds, the design of auxiliary tasks for better representation learning has also become a hot-spot issue in recent studies. In addition to classical auxiliary tasks like depth estimation [61], [65], monocular 2D and 3D detection [11], [71], and 2D lane detection [31], several works also devise schemes for knowledge distillation from cross-modality settings such as monocular learn from stereo [127] and stereo learn from LiDAR [128]. However, this new trend still focuses on experiments on small datasets, requiring further validation and development on large-scale datasets where a large amount of training data may weaken the benefits of such training approaches.…”

Section: Auxiliary Tasksmentioning

confidence: 99%

Vision-Centric BEV Perception: A Survey

Ma¹,

Wang²,

Bai³

et al. 2022

Preprint

View full text Add to dashboard Cite

Vision-centric BEV perception has recently received increased attention from both industry and academia due to its inherent merits, including presenting a natural representation of the world and being fusion-friendly. With the rapid development of deep learning, numerous methods have been proposed to address the vision-centric BEV perception. However, there is no recent survey for this novel and growing research field. To stimulate its future research, this paper presents a comprehensive survey of recent progress of vision-centric BEV perception and its extensions. It collects and organizes the recent knowledge, and gives a systematic review and summary of commonly used algorithms. It also provides in-depth analyses and comparative results on several BEV perception tasks, facilitating the comparisons of future works and inspiring future research directions. Moreover, empirical implementation details are also discussed and shown to benefit the development of related algorithms.

show abstract

“…In recent years, target-detection algorithms have been widely studied in the fields of pedestrian detection [10][11][12][13][14], object tracking [15][16][17], face detection [18,19], stereo images [20][21][22], car detection [23][24][25], defect detection [26,27], semantic detection [28,29], and hyperspectral-anomaly detection [30,31]. However, there are certain limitations in their practical application, particularly due to the problems of poor small-target-detection performance and target occlusion.…”

Section: Introductionmentioning

confidence: 99%

An Efficient Ship-Detection Algorithm Based on the Improved YOLOv5

Wang,

Pan,

et al. 2023

Electronics

View full text Add to dashboard Cite

Aiming to solve the problems of large-scale changes, the dense occlusion of ship targets, and a low detection accuracy caused by challenges in the localization and identification of small targets, this paper proposes a ship target-detection algorithm based on the improved YOLOv5s model. First, in the neck part, a weighted bidirectional feature pyramid network is used from top to bottom and from bottom to top to solve the problem of a large target scale variation. Second, the CNeB2 module is designed to enhance the correlation of coded spatial space, reduce interference from redundant information, and enhance the model’s ability to distinguish dense targets. Finally, the Separated and Enhancement Attention Module attention mechanism is introduced to enhance the proposed model’s ability to identify and locate small targets. The proposed model is verified by extensive experiments on the sea trial dataset. The experimental results show that compared to the YOLOv5 algorithm, the accuracy, recall rate, and mean average precision of the proposed algorithm are increased by 1.3%, 1.2%, and 2%, respectively; meanwhile, the average precision value of the proposed algorithm for the dense occlusion category is increased by 4.5%. In addition, the average precision value of the proposed algorithm for the small target category is increased by 5% compared to the original YOLOv5 algorithm. Moreover, the detection speed of the proposed algorithm is 66.23 f/s, which can meet the requirements for detection speed and ensure high detection accuracy and, thus, realize high-speed and high-precision ship detection.

show abstract

SGM3D: Stereo Guided Monocular 3D Object Detection

Cited by 21 publications

References 36 publications

Recent Advancements in Deep Learning Applications and Methods for Autonomous Navigation -- A Comprehensive Review

Recent Advancements in Deep Learning Applications and Methods for Autonomous Navigation -- A Comprehensive Review

Vision-Centric BEV Perception: A Survey

An Efficient Ship-Detection Algorithm Based on the Improved YOLOv5

Contact Info

Product

Resources

About